OpenAI Releases ChatGPT Images 2.5 with 50% Lower Latency and Sketch Controls

OpenAI has announced ChatGPT Images 2.5, reducing image generation latency by up to 50% while introducing interactive sketch-to-image controls, template libraries, and targeted in-image visual annotations. The release also introduces two API options: `gpt-image-2.5-flare` for high-volume generation and `gpt-image-2.5-sunburst` for fine-tuned editing workflows. Significant latency reductions combined with precise localized editing allow creators and developers to iterate on visual assets much faster without quality degradation over multiple edits. By introducing direct visual annotations and sketch inputs, OpenAI bridges the gap between text prompting and intuitive visual design workflows. Users can invoke the `@Sketch` tool to convert simple hand drawings into detailed artwork, or mark specific image regions to issue localized modification commands while maintaining overall spatial consistency. Additionally, all generated outputs automatically embed C2PA metadata and invisible watermarks to ensure content provenance and guard against unauthorized misuse.